> http://www.mailpure.com/software/decludefilters/anti-av/Anti-A
> V_v1-0-2.zip

Oh.. I've must have missed this.
At the moment I'm testing your filter file. 

Looks like with 1.77beta the END statements doesn't work so I have to
comment them out.
This creates a lot of positives result (adding fortunately only a few
points) that has nothing to do with NDRs caused by MassMailer worms.

At the moment I've not adapted the weights for my HOLD-ON-100 system but I
can see that both catched MM-NDRs and false positives will have the same
points (most of them between 2 to 8 points) So I assume this will create a
lot of false positives.


> Unfortunately there is no good way to separate unknown user 
> or full account NDR's from some others, however you can 
> uncomment the null sender line in this filter to add some 
> points so that any suggestive hit will put it over the top.  
> You can also set up a filter with just the null sender and 
> block (hold) that per domain when outbreaks are particularly 
> bad, but bare in mine that legit bounces can be blocked as well.

In my opinion your filter file can help a little bit but I'me pretty sure
that to solve this problem we need some AND/OR/NOT logic. As I understand
holding a legit NDR does not only violate some RFCs but can create also
seriuos problems because neither sender nor recipient know about the failed
delivery.

Anyone else can see more MassMailer-NDRs then Spam passing the filters?

Scott: Are combo tests a tecnical or a "missing-time" problem or do you
consider them not so usefull?

Markus




---
[This E-mail was scanned for viruses by Declude Virus (http://www.declude.com)]

---
This E-mail came from the Declude.JunkMail mailing list.  To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail".  The archives can be found
at http://www.mail-archive.com.

Reply via email to